Multibiometrics for Human Identification

Contents......Page 6
List of Contributors......Page 8
Preface......Page 14
Introduction......Page 16
Part I: Multimodal and Multisensor Biometric Systems......Page 17
Part III: Hybrid Biometric Systems......Page 18
Part V: Performance of Multibiometric Systems......Page 19
PART I: Multimodal and Multisensor Biometric Systems......Page 22
1.1 Introduction......Page 24
1.2 Related Work......Page 26
1.2.1 3D Ear Recognition......Page 27
1.2.2 2D Face Recognition......Page 28
1.3 3D Ear Modeling and Recognition Using Shape from Shading......Page 29
1.3.1 Linear Shape from Shading......Page 30
1.3.2 Similarity Accumulator......Page 33
1.3.3 3D Model Selection......Page 35
1.4 2D Face Recognition Using Gabor Features......Page 36
1.5 Data Fusion......Page 38
1.6 Experiments and Results......Page 40
1.7 Conclusions and Future Work......Page 42
Appendix......Page 43
References......Page 44
2.1 Introduction......Page 46
2.2 The Mutual Information–Based Approach......Page 48
2.3 The Hypothesis Testing–Based Approach......Page 49
2.4 Time Evolution–Based Bimodal Linear Prediction......Page 50
2.5 Generalized Bimodal Linear Prediction......Page 53
2.6 Audiovisual Feature Pair Selection......Page 55
2.7 Experimental Setup......Page 56
2.8 Experimental Results......Page 58
2.8.1 “System-1” Results of BLPC-1......Page 60
2.8.3 Proposed BLPC-1 versus Baseline Approaches......Page 61
2.9 Discussion......Page 62
2.10 Conclusions......Page 63
References......Page 64
3.1 Introduction......Page 66
3.2 Image Acquisition......Page 68
3.3.2 Central Palm Region Localization......Page 71
3.4 Image Registration......Page 73
3.5.1 Pixel-Level Fusion......Page 77
3.5.2 Matching Score-Level Fusion......Page 79
3.6.1 Baseline Performances......Page 81
3.6.2 Division of Wavelengths......Page 82
3.6.3 Pixel-Level Fusion......Page 83
3.6.4 Matching Score-Level Fusion......Page 85
3.7 Conclusion......Page 86
References......Page 87
4.1 Introduction......Page 89
4.2 Methodology......Page 91
4.2.1 Face Segmentation......Page 92
4.2.2 Blood Vessel Segmentation......Page 93
4.3 Vessel Segmentation Post-Processing......Page 94
4.4 Vascular Network Matching......Page 96
4.4.1 Registration of Vascular Networks......Page 97
4.4.2 Face Pose Estimation......Page 98
4.5 Experiments......Page 99
4.5.1.1 Facial Expression Dataset......Page 100
4.5.1.2 Facial Pose Dataset......Page 101
4.5.2 Experiments on the University of Notre Dame Database......Page 102
4.5.3 Experiments on the University of Arizona Database......Page 103
4.6 Conclusions......Page 104
References......Page 106
PART II: Fusion Methods in Multibiometric Systems......Page 108
5.1 Introduction......Page 110
5.2 Fusion of Multibiometrics......Page 112
5.3 Statistical Modeling of Heterogeneous Biometric Data......Page 114
5.3.1 Copula Theory and Its Implications......Page 116
5.4.1 Log-Likelihood Ratio Test Based on Copulas (LLRT-cop)......Page 118
5.4.3 Performance Analysis......Page 119
5.5 Joint PDF Construction Using Copulas......Page 121
5.5.2 Maximum Likelihood Estimation......Page 123
5.6 Experimental Results......Page 124
5.7 Concluding Remarks......Page 131
References......Page 133
6.1 Introduction......Page 135
6.2.1 Face Recognition Based on the Scale-Invariant 
Feature Transform......Page 138
6.2.2 Fingerprint Matching Based on Minutiae Representation......Page 141
6.3.1 Feature Set Compatibility......Page 144
6.3.3 Feature Selection......Page 145
6.4 Matching Feature Vectors......Page 147
6.5 Experimental Results......Page 149
References......Page 154
7.1 Introduction......Page 158
7.2 Self-adaptive Biometric Systems: A Review......Page 160
7.3.1 The Template Co-update Algorithm......Page 162
7.3.2.1 Terminology......Page 165
7.3.2.3 Gallery Size Increase with Supervisor......Page 166
7.3.2.4 Gallery Size Increase without Supervisor......Page 169
7.3.2.5 False Rejection Rate Modeling......Page 172
7.3.2.6 Ensemble of Matchers......Page 174
7.4.2 Experimental Protocol......Page 176
7.4.3 Results......Page 177
7.5 Discussion......Page 180
7.6 Conclusions......Page 181
References......Page 182
PART III: Hybrid Biometric Systems......Page 186
8.1 Introduction......Page 188
8.2 Structured Light System......Page 189
8.2.1.1 Camera Model......Page 190
8.2.1.2 Projector Model......Page 191
8.2.2 Calibration Method......Page 192
8.2.2.2 Projector Calibration......Page 193
8.2.3.1 Calibration......Page 194
8.2.3.2 Evaluation......Page 195
8.3 3D Human Body Measurement......Page 198
8.3.1.1 Projector-Camera Pair......Page 199
8.3.1.2 Pole Unit......Page 200
8.3.1.4 Measurement Time......Page 201
8.3.2.1 Projector-Camera Pair......Page 202
8.3.2.2 Coordinate System Alignment......Page 203
8.3.3.1 Calibration......Page 204
8.3.3.2 Evaluation......Page 205
8.4 Human Gait Recognition......Page 207
8.4.1 3D Human Body Data......Page 209
8.4.2 3D Human Body Model......Page 210
8.4.3 Model Fitting......Page 211
8.4.4 Gait Reconstruction......Page 212
8.4.5 Feature Matching......Page 213
8.4.6.1 Sensing and Modeling......Page 214
8.4.6.3 Recognition......Page 215
8.5 Conclusion......Page 217
References......Page 218
9.1 Introduction......Page 221
9.1.1 Overview of Chapter......Page 223
9.2.1 Related Work on Fusion of Multiple Features in Activity Recognition......Page 224
9.2.2 Related Work in Neurobiology and NMC......Page 225
9.2.3.2 Shape-Based Methods......Page 226
9.3.1 Dorsal Pathway: Hamiltonian Energy Signatures (HES)......Page 227
9.3.1.1 Application to Gait Analysis......Page 229
9.3.1.3 Advantages......Page 230
9.3.2 Ventral Pathway: Shape-Based Features......Page 231
9.3.3 NMC Integration and Gait Analysis......Page 232
9.3.3.1 Hypothesis Testing......Page 233
9.3.3.2 Bootstrap......Page 234
9.3.3.3 Integration Variants......Page 236
9.4 Experimental Results......Page 237
9.5 Conclusion......Page 240
9.A.1.1 Hamilton’s Variational Principle......Page 241
9.A.1.3 Invariance of HES under Affine Transformations......Page 242
References......Page 246
10.1 Introduction......Page 250
10.2.1 Multiview Tracking......Page 252
10.2.3 Spherical Harmonics......Page 253
10.3.2 State Transition Model......Page 254
10.3.3 Observation Model......Page 255
10.3.4 Tracking Results......Page 257
10.4 Spherical Harmonics and Rotational Invariance......Page 258
10.5.1 Multiview Face Recognition Using Spherical Harmonics Features......Page 262
10.5.3 Comparison of Discrimination Power of Features......Page 263
10.5.4 Comparison of Recognition Performance......Page 266
10.5.5 Analysis of Results......Page 268
10.6 Conclusions and Future Work......Page 269
References......Page 270
11.1 Introduction......Page 273
11.2 Related Work......Page 274
11.3.1 2D+3D Enrollment......Page 275
11.3.2.1 Analytical Skin Reflectance Model......Page 276
11.3.2.2 Bidirectional Relighting......Page 278
11.3.2.3 Distance Metric......Page 280
11.4.1 Bidirectional Relighting......Page 281
11.4.2 Face Recognition......Page 282
11.4.2.1 Databases......Page 283
11.5 Conclusions......Page 286
References......Page 287
PART IV: Databases and Security......Page 290
12.1.1 Multibiometrics......Page 292
12.1.2 Multibiometric Data......Page 293
12.1.3.2 Ear Biometrics......Page 295
12.1.4 On Our New Database......Page 296
12.2 Data Collection......Page 298
12.2.1 Gait Tunnel......Page 299
12.2.1.1 Gait Data......Page 301
12.2.1.3 Semantic Data......Page 304
12.3.1 Gait......Page 307
12.3.3 Score-Based Fusion......Page 311
12.4 Conclusions......Page 313
References......Page 314
13.2 Why Dynamic Security Management?......Page 317
13.2.1 Objectives......Page 319
13.3 Summary of Related Work......Page 320
13.4 Quantifying the Security Level......Page 322
13.5.1 Particle Swarm Optimization......Page 323
13.5.2 Score-Level Combinations......Page 325
13.6.1 NIST BSSR1 Database......Page 326
13.6.2 Iris and Palm Database......Page 327
13.7 Summary and Further Directions......Page 331
References......Page 333
PART V: Performance of Multibiometric Systems......Page 336
14.1 Introduction......Page 338
14.2.1 Related Work......Page 340
14.2.2 Contributions......Page 341
14.3.1 Performance Measurements......Page 342
14.3.2.1 Decomposition of the Area under the ROC Curve......Page 345
14.3.2.2 Fusion Prediction......Page 347
14.3.2.3 Summary......Page 349
14.3.3.1 Prediction Model for Fusion......Page 350
14.4.1 Databases......Page 353
14.4.2 Verification of the Match/Non-match Score Distributions as Mixtures of Gaussian......Page 354
14.4.3 Normalization of Scores......Page 356
14.4.4.1 Prediction for the XM2VTS Database......Page 359
14.4.4.2 Prediction for the Mixed Database......Page 363
14.4.5 Performance of Prediction Model Based on the Neyman-Pearson Theory......Page 366
14.5 Conclusions......Page 374
References......Page 375
15.1 Introduction......Page 378
15.2 Experimental Setup......Page 379
15.3 The Score Mixing Effect......Page 380
15.3.1 Example of Identification Systems......Page 381
15.3.2 Performance of Systems with Randomized Impostor Scores......Page 382
15.4 The Binomial Approximation Effect......Page 383
15.5 The Combination of Score Mixing and Binomial Approximation Effects......Page 386
15.6 T-Normalization......Page 388
15.7 Resampling Methods......Page 391
15.7.1 Resampling Using Genuine Score Neighbors......Page 392
15.7.2 Score Resampling Using Identification Trial Statistics......Page 393
15.7.3 Resampling and T-Normalization......Page 394
15.7.4 Resampling Using nth Order Statistics......Page 395
15.7.5 Justification of Using Identification Trial Statistics......Page 396
15.8.1 Identifciation Models......Page 397
15.8.2 Extreme Value Theory......Page 398
15.8.3 Performance Prediction in Open Set Identification Systems......Page 400
15.9 Conclusion......Page 401
References......Page 402
Color Plates......Page 404